INTERNATIONAL JOURNAL OF CLIMATOLOGY
Int. J. Climatol. 20: 721–732 (2000)
RAINFALL VARIABILITY IN SOUTHERN SPAIN ON DECADAL TO
CENTENNIAL TIME SCALES
F.S. RODRIGOa,*, M.J. ESTEBAN-PARRAb, D. POZO-VÁZQUEZc and Y. CASTRO-DÍEZb
a
Dpto. Fı́sica Aplicada, Uni6ersidad de Almerı́a, E-04120 Almerı́a, Spain
b
Dpto. Fı́sica Aplicada, Uni6ersidad de Granada, E-18071 Granada, Spain
c
Dpto. Fı́sica Aplicada, Uni6ersidad de Jaén, E-23071 Jaén, Spain
ABSTRACT
In this work a long rainfall series in Andalusia (southern Spain) is analysed. Methods of historical climatology were
used to reconstruct a 500-year series from historical sources. Different statistical tools were used to detect and
characterize significant changes in this series. Results indicate rainfall fluctuations, without abrupt changes, in the
following alternating dry and wet phases: 1501–1589 dry, 1590 – 1649 wet, 1650 – 1775 dry, 1776-1937 wet and
1938–1997 dry. Possible causal mechanisms are discussed, emphasizing the important contribution of the North
Atlantic Oscillation (NAO) to rainfall variability in the region. Solar activity is discussed in relation to the Maunder
Minimum period, and finally the past and present are compared. Results indicate that the magnitude of fluctuations
is similar in the past and present. Copyright © 2000 Royal Meteorological Society.
KEY WORDS:
regional rainfall anomalies; Little Ice Age; Maunder Minimum; North Atlantic Oscillation (NAO)
1. INTRODUCTION
The third session of the Conference of the Parties to the UN Framework Convention on Climate Change
in Kyoto, Japan, established the necessity of gathering climatic data on a century scale. Time scales
ranging from a week to centuries are needed to elucidate processes governing climatic phenomena such as
droughts or floods, and to identify the causes of long-term shifts in climate (Schneider, 1998). According
to the Intergovernmental Panel on Climate Change (IPCC) (Houghton et al., 1996), further work is
needed on the ‘systematic collection of long-term instrumental and proxy observations of climate system
variables for the purposes of model testing, assessment of temporal and regional variability and for
detection and attribution studies’ to reduce uncertainties in predicting and detecting future climate
change. The present work is an effort in this sense.
This work is the continuation of a previous paper (Rodrigo et al., 1999) where a 500-year precipitation
record in southern Spain was reconstructed. In the present paper, the objective is to analyse climatic
change phenomena on interannual, decadal and centennial timescales, in search of the mechanisms
governing droughts and floods. The period from 1500 onwards includes natural climate variations
associated with the so-called ‘Little Ice Age’ (LIA), the Maunder Minimum (when a minimum solar
activity was detected), the Industrial Revolution in the 19th century (the beginning of possible anthropogenic influences on climate), as well as the last half of the 20th century, when anthropogenic influence has
become evident (Houghton et al., 1996). The analysis of this long series may help to reveal natural causal
mechanisms of climatic change, and establish an appropriate comparison between past and present. While
most of the palaeoclimatic studies on these time scales focus on temperature (e.g. Mann et al., 1998), the
present study adds other climatic variables, such as rainfall. The region analysed is Andalusia (southern
Spain), where rainfall is governed primarily by the Azores High and the Atlantic Lows with their
associated fronts. Wet/dry periods in Andalusia may be related to cold/warm periods in the Northern
Hemisphere. In this sense, the recent positive temperature anomalies over the North Atlantic and
* Correspondence to: Dpto. Fı́sica Aplicada, Universidad de Almerı́a, E-04120, Almerı́a, Spain; tel.: +34 950 215295; fax: +34 950
215477; e-mail: frodrigo@filabres.ualm.es
Copyright © 2000 Royal Meteorological Society
722
F.S. RODRIGO ET AL.
surrounding land masses and recent dry winter conditions over southern Europe and the Mediterranean
are strongly related to the persistent and exceptionally strong positive phase of the North Atlantic
Oscillation (NAO) index since the early 1970s (Hurrell, 1995). The long-term NAO variability is highly
important to study climatic variations in Europe (Kapala et al., 1998).
Section 2 presents the data series and its statistical properties (a more complete analysis may be found
in Rodrigo et al., 1999). Section 3 is dedicated to identifying possible climatic change situations, applying
various statistical tools and to describing the changes in terms of the known concepts of ‘trends’, ‘abrupt
changes’ or ‘fluctuations’. Section 4 presents the results in relation to possible causal mechanisms, with
special attention on the NAO, the Maunder Minimum period, and the comparison between past and
present climate— that is, before and after the possible onset of anthropogenic influences on climate.
Finally, future research perspectives are outlined.
2. DATA
A previous paper (Rodrigo et al., 1999) has reconstructed rainfall in southern Spain from 1500 to the
present, using original documentary sources covering mainly the southwestern Iberian Peninsula and the
Guadalquivir River Valley (Figure 1). A numeric index was established to characterize rainfall and its
evolution. The main events detected were coded, from severe droughts (I= − 2) to severe floods (I = +2)
Figure 1. Map of the study region
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
SOUTHERN SPAIN RAINFALL VARIABLES
723
on a seasonal timescale. An annual index was formulated to summarize the seasonal indices. This indexing
covered 1501 to 1850. Instrumental data in the region began in Gibraltar in 1791 and lasted until 1997.
By using principal components analysis (PCA), this station was shown to be representative of the region’s
rainfall variability. Ordinal indices were calibrated with the results of other studies of historical climate
and with instrumental precipitation data. The overlapping period 1791–1820 was used to calibrate indices
by means of a linear relationship between qualitative indices and instrumental values with the period
1821–1850 being used to validate the calibration. Figure 2 shows the results, expressed as anomalies with
regard to the average value of the instrumental period 1951–1980 and the 10-year moving average. This
reference period was chosen because it is usually used to normalize the NAO index (Jones et al., 1997).
The average value of total annual rainfall in Gibraltar for 1951–1980 is 805.7 mm, the median is 822.5
mm and the standard deviation (S.D.) is 262.7 mm. Although the rainfall series during the reference
period cannot be considered to be a sample from a normal population, the results of applying different
tests for randomness (runs above and below median, runs up and down, Box–Pierce test) confirm the
hypothesis that the reference period series is random at the 90% confidence level.
The 10-year moving average allows a preliminary view of the temporal evolution of rainfall in the
region. That is, after several years of prevailing dry anomalies during the 16th century, with a rainfall
minimum around 1540, a clear wet period begins at the end of the century, lasting until the mid-17th
century. The 18th century is clearly dry, with a minimum around 1750. Afterwards, rainfall anomalies
increase, reaching a maximum around 1860. From the late 19th century, rainfall progressively decreased,
interrupted only by positive anomalies in the 1960s.
Figure 2. Broken line: rainfall anomalies for the period 1501 – 1997 with regard to the average value of the instrumental period
1951–1980. Solid line: 10-year moving average
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
724
F.S. RODRIGO ET AL.
Table I. Statistics of the annual rainfall anomalies for the period 1501–1997 in Andalusia (southern Spain)a
Parameter
Value (mm)
Average
S.D.
Minimum
Percentile 10%
Percentile 25%
Percentile 50%
Percentile 75%
Percentile 90%
Maximum
Standard skewness
Standard kurtosis
−0.3
257.9
−864.0
−257.7
−106.1
−106.1
186.7
348.6
1025.1
5.4
6.9
a
Anomalies are expressed as deviations from the reference period 1951–1980.
Table I shows summary statistics for the series, including measures of the central tendency, variability
and shape. The standardized skewness and kurtosis can be used to determine whether the samples come
from a normal distribution. Values of these statistics outside the range − 2 to + 2 indicate significant
departures from normality. In this case, the standardized skewness and kurtosis values were not within the
range expected for data from a normal distribution. The Chi-Square and the Kolmogorov–Smirnov tests
also imply that the series did not come from a normal distribution with 99% confidence. Non-parametric
statistics are particularly useful when the data distribution is unknown or not normal (Sneyers, 1992). On
the other hand, some tests that were originally developed on the assumption of a normal distribution
turned out to be quite robust when applied to data in which the distribution deviated considerably from
the Gaussian form. In this sense, the t-test for comparison between means may be considered robust and
it may be applied to data having any arbitrary frequency distribution (Mitchell, 1966). In the following
sections, non-parametric tools or robust tests are applied to study the temporal evolution of rainfall in the
study region.
3. DETECTION AND CHARACTERIZATION OF CHANGES
After climatic reconstruction, the time series was analysed to find behavioural patterns of the variable,
and to compare contemporary with historical rainfall. The objective was to identify possible climatic
change situations. First, the sum of cumulative deviations was calculated. If two subperiods showed a
significant change of their average values, this was reflected in a change of slope in the curve that
represents the cumulative deviations against time. The change point was the year having an extreme value
(Bárdossy and Caspary, 1990). This method has been useful to analyse rainfall and runoff series in which
high variability may mask long-term trends (Mitchell, 1966). However, this method does not identify the
character of the change or the stability of the series before and after the change point, and thus the results
should be considered with caution (Sneyers, 1992). Figure 3 represents the computed sums of cumulative
deviations. The result indicates the most notable changes in the slope of the curve in the years 1589
(absolute minimum of the cumulative deviations), 1649 (maximum in the 17th century), 1775 (minimum
in the 18th century) and 1937 (absolute maximum of the cumulative deviations). Table II shows the
number of years and basic statistics of each. The length of these periods reveals no regular pattern.
Average and median values indicate that the periods 1501–1589, 1650–1775 and 1938–1997 are dry in
relation to the reference period, while 1590 – 1649 and 1776–1937 may be considered wet. The definition
of a wet or dry period affects not only the mean values, but also the extreme values. Thus, in general
terms, dry periods showed the lowest minimum values, and wet periods the highest maximum values.
Skewness and kurtosis coefficients indicate the non-normal character of most of these periods.
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
725
SOUTHERN SPAIN RAINFALL VARIABLES
Figure 3. Sum of cumulative deviations for the southern Spain rainfall, 1501 – 1997
Table II. Periods detected by applying sum of cumulative deviations (values expressed
in mm as deviations from the average of the reference period 1951–1980)
n (years)
Average
S.D.
Median
Minimum
Maximum
Standard skewness
Standard kurtosis
1501–1589
1590–1649
1650–1775
1776–1937
1938–1997
89
−65
219
−106
−712
803
4.2
8.1
60
132
299
178
−409
803
1.4
−0.9
126
−57
233
−106
−864
576
−2.6
5.9
162
54
255
7
−436
1025
4.6
3.4
60
−64
249
−108
−524
730
2.5
1.2
To confirm the existence of these changes, a t-test of difference between the means was applied. This
test involves a comparison of two different periods, before and after a possible change point. To confirm
unsteadiness, the condition td \t must be verified, where td is the defined statistic to verify the null
hypothesis of randomness, and t is Student’s t. Table III shows the results of applying this test to pairs
of successive subperiods within the overall period studied. Each period proved to be significantly different
from the previous one, because in all the cases the alternative hypothesis (statistically significant
difference) was verified. In the second period, a negative sign of the difference suggests an increase in
rainfall, and a positive sign a decrease. The second period (1590–1649) deserves special mention, because
not only the differences between means are statistically significant, but also the S.D.s, as indicated by the
F-test (in this case, a t-test was applied without assuming equal variances). These results are similar to
findings of other authors. Briffa et al. (1990), from tree-ring analyses in Fennoscandia, established a
relatively short cold period between 1570 and 1650, with a warmer period before 1570, the main cold
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
726
F.S. RODRIGO ET AL.
Table III. Comparison between the periods detected, F-test to compare S.D.s and t-test for difference between means
(95% confidence levels)a
Periods
F-test
Confidence interval
for ratio of
variances
Difference between
means (mm)
td
Confidence interval
for difference
between means
(1501–1589)–
(1590–1649)
(1590–1649)–
(1650–1775)
(1650–1775)–
(1776–1937)
(1776–1937)–
(1938–1997)
0.59*
(0.33, 0.85)
−197
−4.36*
(−281, −112)
1.65*
(1.08, 2.61)
+189
4.32*
0.83
(0.60, 1.17)
−111
−3.79*
1.05
(0.67, 1.57)
+118
3.06*
a
(109, 269)
(−168, −53)
(42, 193)
Null hypothesis: equal means and S.D.s (*denotes a statistically significant difference).
period between 1570 and 1620, and normal conditions from 1660 until 1750. In their analysis on values
of accumulated areal ice volume along the Baltic coast of Germany, Koslowski and Glaser (1999) found
strong phases of increased winter severity in 1593–1630 and 1763–1860, coinciding with wet phases in
Andalusia, and decreased winter severity in 1501–1533, 1711–1762 and from 1861 to the present, in
accordance with the dry phases in southern Spain. This coincidence suggests that the LIA was
characterized in the southern Iberian Peninsula by increased rainfall, with the main phase approximately
between 1590 and 1650.
The t-test established different periods (i.e. a change between periods), but, because of the nature of this
test, did not specify the character of these changes. Thus, the next step was to determine the nature of the
changes detected. In general, three types of changes can be distinguished: abrupt change, trend and
fluctuations. The term ‘trend’ refers not only to a linear change, but also to changes with a maximum or
minimum at the extreme points of the series. An abrupt change occurs when a trend is present and a
change point divides the series into two distinct subseries. The dividing point is an abrupt-change point.
This point is found by applying the sequential version of the Mann–Kendall test (Sneyers, 1992), this
non-parametric test being the most appropriate for detecting abrupt changes in climatological series
(Esteban-Parra et al., 1995). The application of this test (Figure 4) showed that the changes detected were
not abrupt, because the graphs of the sequential onward version of the statistical trend test (solid line) and
the sequential backward curve (broken line) did not intersect at the 95% significance level. A decreasing
Figure 4. Mann – Kendall test of the rainfall series, period 1501 – 1997. Solid line: sequential onward version of the statistic trend
test; broken line: sequential backward version. The 95% significance level is shown
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
727
SOUTHERN SPAIN RAINFALL VARIABLES
trend appeared around 1540, an increasing trend that reached its maximum around 1650, and afterwards
lost its significance, and again an increasing trend in the first half of the 20th century—that is, a result
similar to the dry and wet periods detected with the previous method. These trends were not linear
(correlation coefficients and slopes are 0). Over the long term, abrupt changes were not evident.
Therefore, the temporal evolution of rainfall may be characterized as fluctuating, with alternating maxima
and minima, resulting from non-linear increasing and decreasing trends.
A reliable method of studying time series is via a power spectrum. The standard FFT spectrum was
calculated. However, it is an open question whether the results have any physical meaning or are a
mathematical artefact, and comparisons of different results may highlight this problem. The spectrum was
calculated taking the year as the time unit, so that the results were related to interannual and interdecadal
variations. Figure 5 shows the resulting periodogram. Because the lag-one serial correlation coefficient
(r1 =0.12) differed significantly from zero, a null continuum red noise was adopted. Peaks above the 95%
confidence level indicated possible significant periodicities—that is, possible mechanisms responsible for
the main fluctuations in the series. The greatest part of the total variance accumulated at the lowest
frequency, indicating long-term trends, as indicated by the Mann–Kendall test. These long-term trends
may be related to the onset of the main phase of the LIA (period 1590–1649) and perhaps with global
warming during the 20th century, during which decreasing rainfall (since 1930s) has been detected. In fact,
the results of several experiments of high-resolution GCM match the trend during the last years of the
20th century (Houghton et al., 1996). Similar long-term trends masked by other fluctuations have been
found by Mächel et al. (1998) while analysing the ‘centres of action’ above the Atlantic area from 1881
to the present.
The other peaks detected in Andalusian rainfall correspond to periodicities of about 16.7, 7–9, 3.5 and
2.1 years. In relation to the interdecadal variability with a period of 16.7 years, Rodrı́guez-Puebla et al.
(1998) have found a significant oscillation of 16 years in their analysis of Spanish rainfall for the period
1949–1995 on the Mediterranean coast of the Iberian Peninsula, coinciding with the 16-year period
detected in the Southern Oscillation Index (SOI) teleconnection pattern. In fact, an oscillation in the
Pacific Ocean with a time scale of 10 – 20 years and a significant peak of about 18 years were detected in
the spectrum of sea-surface temperature (SST) over the North Pacific, from an integration of the
Hamburg ECHAM+LSG coupled ocean – atmosphere GCM (von Storch, 1994; Robertson, 1996).
Moron et al. (1998) found a significant oscillation in North Atlantic SST of about 13–15 years. Therefore,
this peak could indicate a mid-latitude air – sea interaction in the Atlantic area. Cycles around 7–9 years
and a quasi-biannual oscillation have been found in the analysis of the NAO index (Hurrell and van
Loon, 1997; Appenzeller et al., 1998; Cook et al., 1998). Luterbacher et al. (1999) reconstructed the
monthly NAO index back to AD1675, and found strong decadal to interdecadal variations and
Figure 5. Power spectra of the rainfall anomalies for the period 1501 – 1997. The 95% significance level is shown
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
728
F.S. RODRIGO ET AL.
significant spectral power between 54 and 68 years in the annual mean NAO index. According to these
authors, further studies are needed to clarify whether their results were affected by shortcomings of their
data base and/or reconstruction method. Finally, the SST of the Mediterranean from 1950 to 1988 shows
a periodicity of about 3.1 years (Makrogiannis and Sahsamanoglou, 1990), and the main values of
pressure of the Azores High show a periodicity of about 4 years (Sahsamanoglou, 1990). Therefore, it is
plausible to associate these periodicities with the Atlantic ‘centres of action’, the NAO and ocean–
atmosphere interactions.
4. DISCUSSION
The periodicities found closely agree with the results of spectral analysis applied to the behaviour of
Atlantic action centres and the NAO index, this index usually being defined for the winter season
(Hurrell, 1995; Jones et al., 1997). In a previous paper (Esteban-Parra et al., 1998), the correlation
coefficients between the NAO index computed by Hurrell (1995) and a series of precipitation in Spain
were calculated. The results, significant at 95% confidence level, for the study region gave − 0.55 for
annual rainfall, and − 0.67 for winter rainfall. From a Canonical Correlation Analysis, Zorita et al.
(1992) found similar correlations and established that the NAO pattern is the most important atmospheric
phenomenon in the Atlantic area associated with Iberian winter rainfall. Therefore, the high-index values
can be related to low annual and winter precipitation (drought), and the low-index values to intense
rainfall and so floods. This inverse relationship appears because westerlies, associated with high-index
values, are associated with an intensification of the high pressure system over the western Iberian
Peninsula, producing blocking situations over the study area. By contrast, low-index values, and
intensified meridional circulation, shift the Atlantic cyclone track southwards, invading the Iberian
Peninsula, and causing intense rainfall and so floods. Table IV shows the NAO winter index values
corresponding to very dry and very wet winters in Gibraltar from the period 1864–1997. Very dry and wet
winters are defined according the percentile distribution of rainfall at this station (B P10 and \ P90,
respectively). Extreme NAO index values are defined as those outside the interval (x̄− s, x̄+ s) where x̄
is the mean value and s the S.D. of the NAO winter index. The mean value was −0.03 and the S.D. of
the NAO index was 1.8. Very wet winters were associated with negative values, except in 1912, 1933 and
Table IV. NAO winter index corresponding to very wet and very dry winters in
Gibraltar for the period 1864–1997a
Very dry winters
Very wet winters
Year
Rainfall
(mm)
NAO index
Year
Rainfall
(mm)
NAO index
1869
1875
1906
1919
1925
1931
1943
1957
1961
1973
1981
1983
1994
1995
141
165
99
170
146
112
140
137
110
159
15
139
146
44
+1.46
−1.50
+1.80
−0.97
+2.12
−0.34
+1.24
+1.27
+1.56
+2.24
+1.77
+4.13
+2.03
+3.37
1881
1886
1895
1912
1918
1933
1959
1963
1964
1969
1977
1990
708
661
680
602
762
672
699
931
628
757
639
798
−3.87
−1.28
−4.01
+0.06
−0.97
+0.07
−0.55
−3.66
−2.94
−4.90
−2.28
+3.66
Very wet winters defined as those with rainfall \percentile 90% ( = 598 mm), and very dry
winters defined as those with rainfall Bpercentile 10% ( =180 mm) of the period 1864–1997.
a
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
SOUTHERN SPAIN RAINFALL VARIABLES
729
1990; very dry winters were associated with positive values, except in 1875, 1919 and 1931. However,
exceptions were associated with low index values, very close to the mean value or within the normal
interval (x̄− s, x̄+ s). Only 1990 presents an apparent contradictory situation. That is, 50% of the very
wet winters were associated with an extreme negative NAO values, and 40% of the very dry winters were
associated with an extreme positive NAO index. In this sense, a recent reconstruction of the NAO index
for the past 350 years from ice cores in Greenland (Appenzeller et al., 1998) has shown the highest values
of the NAO index occur around 1695, in the early 18th century and 20th century, and the lowest values
around 1680, 1880 and 1960. According to Koslowski and Glaser (1999), the identification of variations
in ice winter severity along the German Baltic coast indicate the probable state of the NAO: periods of
low ice production (1501 – 1553, 1711 – 1762) are strongly correlated with strong westerlies, and heavy ice
production (1593– 1630, 1763 – 1880) with weak westerlies. In general terms, these results coincide with the
appearance, respectively, of dry and wet periods in the Andalusian data. The NAO is responsible for
generating systematic large-scale anomalies in wind speed and latent and sensible heat fluxes, and,
therefore, is related to feedback mechanisms between ocean and atmosphere. According to Hurrell (1996)
the mechanisms of hemispheric warming amplifies the NAO in the Atlantic sector. Moron et al. (1998)
found a 7.5 year oscillation in North Atlantic SST, and related this oscillation to the interannual
variability of the NAO index, although these authors recognize that the interaction of such an SST
oscillation with the atmosphere above deserves further exploration. Other features require more thorough
research, as, for example, the statistical comparison with other reconstructions, in particular the recent
reconstructions of the NAO index (Appenzeller et al., 1998; Cook et al., 1998; Luterbacher et al., 1999).
In relation to the Maunder Minimum (1645 –1715), a period inside the LIA in which the solar activity
decreased notably, many studies on climatic change have emphasized solar activity as a cause of climatic
change (Eddy, 1976; Reid, 1993). In addition, to support these ideas, spectral analysis of many
meteorological variables reveals significant peaks having periodicities similar to those of solar activity
(11, 22 years). In the present case, the Maunder Minimum is included in one of the dry periods detected,
and does not seem exceptional within the overall series. The evolution of rainfall is highly variable, with
severe droughts (1664, 1683, 1689, 1704, 1711) and floods (1646, 1649, 1684, 1692). This high variability
has been detected not only in southern Spain, but also in other parts of Europe, such as in Italy (Camuffo
and Enzi, 1994). The mean value of rainfall anomalies in Andalusia for this period is −25 mm and the
S.D. is 239 mm, values very similar to those of the entire series. When this period is compared with other
periods of similar length, such as the immediately preceding period, 1575–1644, the period immediately
afterward, 1716– 1785, and the last 70 years in the 20th century, the following results were found: the
t-test of difference between means established a significant difference with respect to the period
1575–1644. This difference is a consequence of the wet character of the period from the last decades of
the 16th century to the first half of the 17th century—that is, the interval that has been characterized as
the main phase of the LIA in Andalusia. The difference with respect to the period 1716–1875 is not
significant, and the result is similar when compared with the instrumental values for the period
1927–1997. In no case is there a statistically significant difference between the two S.D.s; thus, in
comparing means, it is assumed that the variances of the two samples are equal. Consequently, the
Maunder Minimum period was not exceptional in the context of the overall record, and, therefore, it
cannot be affirmed that the decreasing trend of solar activity during the Maunder Minimum had a
notable influence on Andalusian rainfall. A similar result was reported after analysing temperature
reconstructions from tree rings in western and southern Europe (Serre-Bachet, 1994). According to some
authors (Lansberg, 1984; Legrand et al., 1990), the Maunder Minimum did not affect the LIA climate.
However, other authors (Appenzeller et al., 1998; Koslowski and Glaser, 1999) have detected a clear
Minimum Maunder signal in their analysis from ice cores in Greenland (highest value of the proxy index
around 1695) and ice volume along the German Baltic coast (increased winter severity in 1655–1710).
This is a major discrepancy with regard to Andalusian data. A possible explanation is that only a few
extremely warm and cool decades were synchronous over the entire continent (Jones and Bradley, 1992).
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
730
F.S. RODRIGO ET AL.
Table V. Comparison between dry and wet periods detected, F-test to compare S.D.s and t-test for difference
between means (95% confidence levels)a
Periods
F-test
Confidence interval
for ratio of
variances
(1501–1589)–
(1938–1997)
(1650–1775)–
(1938–1997)
(1590–1659)–
(1776–1937)
0.78
(0.48, 1.23)
0.87
1.38
a
Difference between
means (mm)
td
Confidence interval
for difference
between means
−1
−0.02
(−77, 76)
(0.55, 1.33)
+7
+0.18
(−67, 80)
(0.92, 2.15)
+78
1.94
(−1, 158)
Null hypothesis: equal means and S.D.s (*denotes a statistically significant difference).
Perhaps solar activity is more influential at high latitudes, or on hemispheric or global scales. However,
recent analyses from the measurement of total solar irradiance, compiled by five independent space-based
radiometers since 1978 – 1996, indicate that total irradiance forcing is unlikely to be the cause of global
warming in the last decade (Fröhlich and Lean, 1998). According to Tol and Vellinga (1998), from a
statistical analysis of the influence of the sun on climatic change, the solar hypothesis loses plausibility.
Now it is possible to compare historical and modern periods. Table V shows the t-test results of the
difference between means applied to compare dry historical periods 1501–1589 and 1650–1775 with the
recent dry period 1938 – 1997, and the wet periods 1590–1649 and 1776–1937. The results show no
significant difference between historical and modern periods—that is, fluctuations in the 20th century
precipitation in southern Spain are similar to those in the past, at least with regard to the mean value.
Whether these variations are caused by natural or anthropogenic forcing remains an open question. In
any case, the results appear to indicate that the magnitude of the changes is similar in both the past and
the present.
5. CONCLUSIONS
The main conclusions of this work are the following:
(i) Rainfall in Andalusia (southern Spain) shows a fluctuating time evolution, with alternate dry and wet
periods. Differences between mean values of these periods (1501–1589 dry, 1590–1649 wet, 1650–
1775 dry, 1776 – 1937 wet and 1938 – 1997 dry) are statistically significant. With slight differences,
these phases coincide with other results found in the literature. Abrupt changes over the long term
were not detected.
(ii) The main phase of the LIA corresponds to the period 1590–1649, characterized by wet conditions,
with higher flooding frequency.
(iii) Spectral analysis shows that the main periodicities in the data series correspond to fluctuations of
about 16.7, 7 – 9, 3.5 and 2.1 years. These fluctuations are superimposed to non-linear long-term
trends.
(iv) Among the possible causal mechanisms of rainfall fluctuations in the region, the NAO behaviour is
the most remarkable, and in particular extreme values of the NAO index are related to droughts
(positive extreme NAO) and floods (negative extreme NAO).
(v) The Maunder Minimum (1645 – 1715) does not appear to be exceptional in the context of the entire
series, indicating that solar activity is unlikely to be the cause of climatic changes in the study region.
(vi) 20th century rainfall anomalies show a behaviour similar to that of other periods in the past.
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
SOUTHERN SPAIN RAINFALL VARIABLES
731
The close relationship between extreme NAO values and precipitation anomalies over southern Spain
could be used to reconstruct a longer record of the NAO index, than is now available, or, at least, to
validate the existing reconstructions. This is an objective of future work. The possibility of monitoring the
long-term NAO variability over an extended period when human influence is negligible may have
implications for comparisons between the past and the present, and, therefore, for climate predictions.
ACKNOWLEDGEMENTS
This work has been carried out under the Research Project CLI98-0930-C02-01 (CICYT), Spanish
Interministerial Commission for Science and Technology.
REFERENCES
Appenzeller C, Stocker TF, Anklin M. 1998. North Atlantic Oscillation dynamics recorded in Greenland ice cores. Science 282:
446 – 449.
Bárdossy A, Caspary HJ. 1990. Detection of climate change in Europe by analysing European atmospheric circulation patterns from
1881 to 1989. Theoretical and Applied Climatology 42: 155 – 167.
Briffa KR, Bartholin TS, Eckstein D, Jones PD, Karlen W, Schweingruber FH, Zetterberg P. 1990. A 1400-year tree-ring record
of summer temperatures in Fennoscandia. Nature 346: 434 – 439.
Camuffo D, Enzi S. 1994. The climate of Italy from 1675 to 1715. In Climatic Trends and Anomalies in Europe 1675 – 1715, Frenzel
B, Pfister C, Glaser B (eds). Gustav Fischer Verlag: Stuttgart.
Cook ER, D’Arrigo RD, Briffa KR. 1998. A reconstruction of the North Atlantic Oscillation using tree-ring chronologies from
North America and Europe. The Holocene 8: 9–17.
Eddy JA. 1976. The Maunder Minimum. Science 192: 1189 – 1202.
Esteban-Parra MJ, Rodrigo FS, Castro-Dı́ez T. 1995. Temperature trends and change points in the Northern Spanish Plateau during
the last 100 years. International Journal of Climatology 15: 1031 – 1042.
Esteban-Parra MJ, Rodrigo FS, Castro-Dı́ez Y. 1998. Spatial and temporal patterns of precipitation in Spain for the period
1880 – 1992. International Journal of Climatology 18: 1557 – 1574.
Fröhlich C, Lean J. 1998. The Sun’s total irradiance: cycles, trends and related climate change uncertainties since 1976. Geophysical
Research Letters 25: 4377–4380.
Houghton JT, Meira Filho LG, Callander BA, Harris N, Kattenberg A, Maskell K (eds). 1996. Climate Change 1995. The Science
of Climate Change. Cambridge University Press, IPCC: Cambridge.
Hurrell JW. 1995. Decadal trends in the North Atlantic Oscillation and relationships to regional temperature and precipitation.
Science 269: 676 – 679.
Hurrell JW. 1996. Influence of variations in extratropical wintertime teleconnections on Northern Hemisphere temperature.
Geophysical Research Letters 23: 665–668.
Hurrell JW, van Loon H. 1997. Decadal variations in climate associated with the North Atlantic Oscillation. Climate Change
Proceedings International Workshop on Climatic Change at High Ele6ation Sites, Wengen, Switzerland.
Jones PD, Bradley RS. 1992. Climatic variations over the last 500 years. In Climate since A.D. 1500, Bradley RS, Jones PD (eds).
Routledge: London and New York; 649–665.
Jones PD, Jónsson T, Wheeler D. 1997. Extension to the North Atlantic Oscillation using early instrumental pressure observations
from Gibraltar and South-West Iceland. International Journal of Climatology 17: 1433 – 1450.
Kapala A, Mächel H, Flohn H. 1998. Behaviour of the centres of action above the Atlantic since 1881. Part II: associations with
regional climate anomalies. International Journal of Climatology 18: 23 – 36.
Koslowski G, Glaser R. 1999. Variations in reconstructed ice winter severity in the Western Baltic from 1501 to 1995, and their
implications for the North Atlantic Oscillation. Climatic Change 41: 175 – 191.
Lansberg HE. 1984. Comments on Wolf’s notes for the Maunder Minimum of sunspots and retraction of some earlier claims.
Technical Note BN-1019, Institute for Physical Science and Technology, University of Maryland.
Legrand JP, LeGoff M, Mazudier C. 1990. Is there a relationship between the Maunder Minimum and the climatic change observed
during the XVIIth century? In Ad6ances in Geosciences, Schröeder W (ed.). IAGA: Washington.
Luterbacher J, Schmutz C, Gyalistras D, Xoplaki E, Wanner H. 1999. Reconstruction of monthly NAO and EU indices back to AD
1675. Geophysical Research Letters 26: 2745–2748.
Mächel H, Kapala A, Flohn H. 1998. Behaviour of the centres of action above the Atlantic since 1881. Part I: Characteristics of
seasonal and interannual variability. International Journal of Climatology 18: 1 – 22.
Makrogiannis TJ, Sahsamanoglou CS. 1990. Time variation of the mean level pressure over the major Mediterranean area.
Theoretical and Applied Climatology 41: 149–156.
Mann M, Bradley RS, Hughes MK. 1998. Global scale temperature patterns and climate forcing over the past six centuries. Nature
392: 779 – 788.
Mitchell JM. 1966. Climatic Change. WMO Technical Note 79, Geneva.
Moron V, Vautard R, Ghil M. 1998. Trends, interdecadal and interannual oscillations in global sea-surface temperatures. Climate
Dynamics 14: 545 –569.
Reid GC. 1993. Do solar variations change climate? EOS 74: 23 – 24.
Robertson AW. 1996. Interdecadal variability over the North Pacific in a multi-century climate simulation. Climate Dynamics 12:
227 – 241.
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)
732
F.S. RODRIGO ET AL.
Rodrigo FS, Esteban-Parra MJ, Pozo-Vázquez D, Castro-Dı́ez Y. 1999. A 500-year precipitation record in Southern Spain.
International Journal of Climatology 19: 1233–1253.
Rodrı́guez-Puebla C, Encinas AH, Nieto S, Garmendia J. 1998. Spatial and temporal patterns of annual precipitation variability
over the Iberian Peninsula. International Journal of Climatology 18: 299 – 316.
Sahsamanoglou HS. 1990. Contribution to the study of action centres in the North Atlantic. International Journal of Climatology
10: 247 – 261.
Schneider SH. 1998. Kyoto protocol: the unfinished agenda. Climatic Change 39: 1 – 21.
Serre-Bachet F. 1994. Annual and summer mean temperature reconstructions from tree-rings in western and southern Europe since
A.D. 1500 with special reference to the late Maunder Minimum. In Climatic Trends and Anomalies in Europe 1675 – 1715, Frenzel
B, Pfister C, Glaser B (eds). Gustav Fischer Verlag: Stuttgart.
Sneyers R. 1992. On the use of statistical analysis for the objective determination of climatic change. Meteorol Zeitschrift 1:
247 – 256.
Tol RSJ, Vellinga P. 1998. Climate change, the enhanced greenhouse effect and the influence of the Sun: a statistical analysis.
Theoretical and Applied Climatology 61: 1–7.
von Storch JS. 1994. Interdecadal variability in a global coupled model. Tellus 46A: 419 – 432.
Zorita E, Kharin V, von Storch H. 1992. The atmospheric circulation and sea surface temperature in the North Atlantic Area in
winter: their interaction and relevance for Iberian precipitation. Journal of Climate 5: 1097 – 1108.
.
Copyright © 2000 Royal Meteorological Society
Int. J. Climatol. 20: 721 – 732 (2000)